Article: Multi-spectral remote sensing image land classification algorithm for unmanned aerial vehicles targeting transmission line corridors Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.38 pp.22 - 41 Abstract: Based on unmanned aerial vehicle multispectral remote sensing technology, this paper proposes a land cover classification algorithm for transmission line corridors to improve the classification accuracy and efficiency in power safety monitoring. By using drones to collect multispectral images, geometric correction is combined with median filtering and wavelet transform denoising to enhance image quality, and spectral and texture features are extracted. A multi-level segmentation system is constructed and classification thresholds are set. A convolutional neural network (CNN) with Softmax classifier and N-P criterion is used to achieve accurate land cover classification. After using the method proposed in this article to classify ground objects in multispectral remote sensing images, the F1 value is as high as 0.97 and the AUC value is as high as 0.936, indicating high classification accuracy and high effectiveness and application performance. Inderscience Publishers - linking academia, business and industry through research

Title: Multi-spectral remote sensing image land classification algorithm for unmanned aerial vehicles targeting transmission line corridors

Authors: Chuntian Ma; Dan Li; Wei Hu; Weidong Liu; Guozhu Yang; Heping Wang

Addresses: State Grid Electric Power Space Technology Company Limited, Beijing, 102211, China ' State Grid Electric Power Space Technology Company Limited, Beijing, 102211, China ' State Grid Electric Power Space Technology Company Limited, Beijing, 102211, China ' State Grid Electric Power Space Technology Company Limited, Beijing, 102211, China ' State Grid Electric Power Space Technology Company Limited, Beijing, 102211, China ' State Grid Electric Power Space Technology Company Limited, Beijing, 102211, China

Abstract: Based on unmanned aerial vehicle multispectral remote sensing technology, this paper proposes a land cover classification algorithm for transmission line corridors to improve the classification accuracy and efficiency in power safety monitoring. By using drones to collect multispectral images, geometric correction is combined with median filtering and wavelet transform denoising to enhance image quality, and spectral and texture features are extracted. A multi-level segmentation system is constructed and classification thresholds are set. A convolutional neural network (CNN) with Softmax classifier and N-P criterion is used to achieve accurate land cover classification. After using the method proposed in this article to classify ground objects in multispectral remote sensing images, the F1 value is as high as 0.97 and the AUC value is as high as 0.936, indicating high classification accuracy and high effectiveness and application performance.

Keywords: transmission line corridor; land cover classification; UAV; multi-spectral remote sensing; convolutional neural network; CNN.

DOI: 10.1504/IJICT.2025.149294

International Journal of Information and Communication Technology, 2025 Vol.26 No.38, pp.22 - 41

Received: 02 Jul 2025
Accepted: 18 Aug 2025
Published online: 22 Oct 2025 *

Title: Multi-spectral remote sensing image land classification algorithm for unmanned aerial vehicles targeting transmission line corridors

Keep up-to-date